Authors retract paper “confirming” that narcolepsy is an autoimmune disease

stmA group of researchers at Stanford and elsewhere is retracting a 2013 paper that another scientist told Nature was “one of the biggest things to happen in the narcolepsy field for some time.”

The Science Translational Medicine paper caused a buzz because it claimed to show that narcolepsy was an autoimmune disease. Here’s the notice:

A.K. De la Herrán-Arita and colleagues retract their Research Article “CD4+ T Cell Autoimmunity to Hypocretin/Orexin and Cross-Reactivity to a 2009 H1N1 Influenza A Epitope in Narcolepsy,” which appeared in the 18 December 2013 issue of Science Translational Medicine. The researchers report that they have been unable to reproduce the paper’s key findings. Specifically, they could not demonstrate a differential Enzyme-Linked ImmunoSpot response of CD4+ T cells from patients with narcolepsy compared to those from normal controls after exposure to the hypocretin peptides HCRT56–68 or HCRT87–99. Because the validity of the conclusions reported in the study cannot be confirmed, they are retracting the article.

Here’s the original abstract:

Narcolepsy, a disorder strongly associated with human leukocyte antigen (HLA)–DQA1*01:02/DQB1*06:02 (DQ0602), is characterized by excessive daytime sleepiness, cataplexy, and rapid eye movement sleep abnormalities. It is caused by the loss of ~70,000 posterior hypothalamic neurons that produce the wake-promoting neuropeptide hypocretin (HCRT) (orexin). We identified two DQ0602-binding HCRT epitopes, HCRT56–68 and HCRT87–99, that activated a subpopulation of CD4+ T cells in narcolepsy patients but not in DQ0602-positive healthy control subjects. Because of the established association of narcolepsy with the 2009 H1N1 influenza A strain (pH1N1), we administered a seasonal influenza vaccine (containing pH1N1) to patients with narcolepsy and found an increased frequency of circulating HCRT56–68– and HCRT87–99–reactive T cells. We also identified a hemagglutinin (HA) pHA1 epitope specific to the 2009 H1N1 strain, pHA1275–287, with homology to HCRT56–68 and HCRT87–99. In vitro stimulation of narcolepsy CD4+ T cells with pH1N1 proteins or pHA1275–287 increased the frequency of HCRT56–68– and HCRT87–99–reactive T cells. Our data indicate the presence of CD4+ T cells that are reactive to HCRT in narcolepsy patients and possible molecular mimicry between HCRT and a similar epitope in influenza pH1N1, pHA1275–287.

The paper has been cited nine times, according to Thomson Scientific’s Web of Knowledge.

We asked the paper’s corresponding authors, Stanford’s Emmanuel Mignot and Elizabeth Mellins, how the issues with the findings came to light. Mignot tells us:

We were just continuing our work based on the finding, trying to establish it as a diagnostic test, but could not replicate it.  No other work is affected, and in fact the DQ binding studies of that article are perfectly fine.  Only the [Enzyme-Linked ImmunoSpot (ELISPOT)] results are in question.

DQ refers to a type of protein on immune system cells, while the ELISPOT assay measures certain kinds of immune system activity.

Mellins tells Retraction Watch:

All the work in question was done in the Mignot lab and then could not be repeated in the Mignot lab. As you can imagine, it is important to me that this be known because I am an immunologist and the work at issue could be mistakenly attributed to my lab.

An aside: A heads-up about this retraction actually appeared online last week, in some versions of the Science Translational Medicine online table of contents. That headline wasn’t linked to anything, however, and the journal tells us it was a snafu.

7 thoughts on “Authors retract paper “confirming” that narcolepsy is an autoimmune disease”

  1. It’s strange that neither the co-authors or the reviewers noticed the similarities between the ELISpot images. E.g. Fig 1B, HCRT25-39, control 58 and patient 44 are identical.

    1. Excellent and important work Sci-Hi!

      Following your hints, here is what I can also see

      Fig. 1 middle right plate (28 SFU) is the same as Fig. 3 lower 2nd from right (43 SFU)

      Fig. 1 lower right plate (17) is the same as Fig. 3 lower panel, lower right (26)

      Fig. 2 upper panel, top row, 3rd from left (24) is the MIRROR IMAGE through the X axis to Fig. 4 top right (35)

      Did I miss any?

      Mirror images do not happen by accident. It is also nothing short of remarkable how the SFU numbers change for the same reused plates.

      Most of the authors are clearly innocent of any subterfuge since they tried to replicate and repeatedly failed. Those are to be congratulated for their persistence and honesty. But, given that this is an incredibly sensitive topic affecting some persistently ill people, there can be no place in biomedical research for the scheming sociopath(s) who did the image manipulations.

      1. @Scrutineer: It should be enough especially since there are mirror images. However, if you are a collector there are some issues in the supplemental data as well.

        1. @Sci-Hi, Ooh you are a tease!

          If you have the naughtiness noted down, might you spare me the effort to exhaustively compare the supplementary figures with each other and the main paper? The comparative effort expands factorially does it not?

          Weight of evidence can be very important, so all known duplications must be presented. I’ll have another go if you can’t for any reason. This is (or, er, was) a very, very important paper, and truth must out. And you seem to be well ahead of the rest of us in that regard.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.